Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Towards Multi-Task Generative-AI Edge Services with an Attention-based Diffusion DRL Approach

Yaju Liu, Xi Lin, Siyuan Li, Gaolei Li, Qinghua Mao, Jianhua Li

TL;DR

The paper addresses the challenge of selecting appropriate AIGC services for heterogeneous, multi-type tasks in resource-constrained edge environments. It introduces ADSAC, a diffusion-policy Soft Actor-Critic framework augmented with an attention mechanism to capture long-range dependencies in high-dimensional state representations. Key contributions include formulating the edge-based AIGC selection as an MDP, designing a DDPM-based policy within SAC, and showing empirically that ADSAC improves user utility while reducing server crashes compared to several DRL baselines and heuristics. The approach offers a scalable, robust solution for dynamic edge-AIGC environments, enabling efficient, low-latency content generation under diverse demands.

Abstract

As an emerging paradigm of content creation, AI-Generated Content (AIGC) has been widely adopted by a large number of edge end users. However, the requests for generated content from AIGC users have obvious diversity, and there remains a notable lack of research addressing the variance in user demands for AIGC services. This gap underscores a critical need for suitable AIGC service selection mechanisms satisfying various AIGC user requirements under resource-constrained edge environments. To address this challenge, this paper proposes a novel Attention-based Diffusion Soft Actor-Critic (ADSAC) algorithm to select the appropriate AIGC model in response to heterogeneous AIGC user requests. Specifically, the ADSAC algorithm integrates a diffusion model as the policy network in the off-policy reinforcement learning (RL) framework, to capture the intricate relationships between the characteristics of AIGC tasks and the integrated edge network states. Furthermore, an attention mechanism is utilized to harness the contextual long-range dependencies present in state feature vectors, enhancing the decision-making process. Extensive experiments validate the effectiveness of our algorithm in enhancing the overall user utility and reducing the crash rate of servers. Compared to the existing methods, the proposed ADSAC algorithm outperforms existing methods, reducing the overall user utility loss and the server crash rate by at least 58.3% and 58.4%, respectively. These results demonstrate our ADSAC algorithm is a robust solution to the challenges of diverse and dynamic user requirements in edge-based AIGC application environments.

Towards Multi-Task Generative-AI Edge Services with an Attention-based Diffusion DRL Approach

TL;DR

The paper addresses the challenge of selecting appropriate AIGC services for heterogeneous, multi-type tasks in resource-constrained edge environments. It introduces ADSAC, a diffusion-policy Soft Actor-Critic framework augmented with an attention mechanism to capture long-range dependencies in high-dimensional state representations. Key contributions include formulating the edge-based AIGC selection as an MDP, designing a DDPM-based policy within SAC, and showing empirically that ADSAC improves user utility while reducing server crashes compared to several DRL baselines and heuristics. The approach offers a scalable, robust solution for dynamic edge-AIGC environments, enabling efficient, low-latency content generation under diverse demands.

Abstract

As an emerging paradigm of content creation, AI-Generated Content (AIGC) has been widely adopted by a large number of edge end users. However, the requests for generated content from AIGC users have obvious diversity, and there remains a notable lack of research addressing the variance in user demands for AIGC services. This gap underscores a critical need for suitable AIGC service selection mechanisms satisfying various AIGC user requirements under resource-constrained edge environments. To address this challenge, this paper proposes a novel Attention-based Diffusion Soft Actor-Critic (ADSAC) algorithm to select the appropriate AIGC model in response to heterogeneous AIGC user requests. Specifically, the ADSAC algorithm integrates a diffusion model as the policy network in the off-policy reinforcement learning (RL) framework, to capture the intricate relationships between the characteristics of AIGC tasks and the integrated edge network states. Furthermore, an attention mechanism is utilized to harness the contextual long-range dependencies present in state feature vectors, enhancing the decision-making process. Extensive experiments validate the effectiveness of our algorithm in enhancing the overall user utility and reducing the crash rate of servers. Compared to the existing methods, the proposed ADSAC algorithm outperforms existing methods, reducing the overall user utility loss and the server crash rate by at least 58.3% and 58.4%, respectively. These results demonstrate our ADSAC algorithm is a robust solution to the challenges of diverse and dynamic user requirements in edge-based AIGC application environments.
Paper Structure (10 sections, 14 equations, 4 figures, 1 table)

This paper contains 10 sections, 14 equations, 4 figures, 1 table.

Table of Contents

  1. Introduction
  2. Related Works
  3. Problem Formulation
  4. Attention-based Diffusion Soft Actor-Critic for AIGC Service Selection
  5. Denoising Diffusion Probabilistic Model
  6. Attention-based Diffusion Soft Actor-Critic
  7. Experiments
  8. Experimental Setup
  9. Performance and Analysis
  10. Conclusion

Figures (4)

  • Figure 1: Edge computing-supported AIGC service architecture for multi-type task scenario.
  • Figure 2: Structure of DDPM as policy in reinforcement learning algorithm.
  • Figure 3: Comparison of test reward curves of the ADSAC algorithm with popular value-based reinforcement learning algorithms, policy-based reinforcement learning algorithms, and heuristic algorithms, respectively.
  • Figure 4: Test rewards over different task arrival rates.